Chapter 5

Flickers and Shadows

More Ways to Find Planets

Since faint planets are hard to see next to bright stars, astronomers have had to come up with clever ways to unveil them. The Doppler technique—using spectral line shifts to trace the subtle dance of stars as planets tug on them—has been the most successful in the first fifteen years. But two other methods have also reached maturity—and are paying off handsomely. Both depend on finding chance alignments of celestial objects through brightness changes of stars.

The first technique exploits a remarkable property of gravity that Albert Einstein discovered: its ability to bend light, thus to magnify the brightness of a distant star temporarily when a nearer star happens to cross our line of sight to the former. If the nearby star harbors a planet, the planet’s gravity causes an extra blip, betraying its presence. The second method relies on a phenomenon scientists have known about for nearly four centuries. Every once in a while, we see Venus and Mercury cross the Sun, appearing as a little black dot against the bright solar disk. These mini-eclipses, called transits, occur when one of these inner two planets passes precisely between the Sun and the Earth. Similarly, if the orbit of an extrasolar planet is aligned exactly edge-on from our vantage point, we can see the star’s brightness dipping ever so slightly each time the planet passes in front of it over the course of its orbit.

Warped Light

By day, Jennie McCormick seems like an ordinary woman living in an eastern suburb of Auckland, New Zealand. Smart but stubborn, she quit school at sixteen and helped train and ride racehorses for a few years before getting married and having two sons. Now forty-five, she works full-time as a personal assistant to the general manager of a storage-bin production company. But at night, she transforms into an avid stargazer. Keenly interested in astronomy ever since she was a little girl, she bought her first telescope as a young mother. But it was only in 2000 that she got serious about “making a contribution.” She contacted the Center for Backyard Astronomy, a worldwide network monitoring a class of temperamental stars known as cataclysmic variables. With their help, she obtained a CCD (charge-coupled device) camera for her 10-inch Meade LX200 telescope. The telescope, while puny by professional standards, was fairly high tech. It tracked celestial objects well and could be operated remotely by computer from the warmth of a home office. That made it easy for McCormick to log many hours of observing while making dinner or ironing clothes. She said she would rather stay home and scan the heavens than go out with her friends. “It’s more than a hobby, it’s a lifestyle.” On April 18, 2005, with the sky perfectly clear, McCormick got ready for what she expected would be another good but routine night of observing. There was a pesky little problem, however: the top two targets on her list were both obscured, one by a palm tree and the other by a TV aerial. Determined not to waste such a beautiful night, she looked around for other possibilities. Her e-mail inbox contained a message from someone at Ohio State University—probably a graduate student, she assumed—requesting observations of a “microlensing event.” McCormick had no idea what that meant, but the target was in the direction of the Milky Way’s bulge, high in the sky above Auckland, and easy to acquire. She decided to give it a try.

Little did McCormick know that she, a Kiwi mother with no formal scientific training, was treading on the legacy of Albert Einstein, possibly the most celebrated scientist of all time. In his general theory of relativity, completed in 1915, Einstein proposed a whole new theory of gravity. Instead of the Newtonian idea of gravity as an attractive force, he conceptualized gravity as geometry: a massive object warps the fabric of space-time around it. That means light, instead of traveling in a straight line, takes a curved path in its vicinity. Einstein’s equations predicted by just how much the light’s path would bend.

The stunning confirmation came four years later. A total solar eclipse was to take place on May 29, 1919. Conveniently, it would occur in front of a rich cluster of stars known as the Hyades, offering an excellent opportunity to measure any defection of starlight by the Sun’s gravity. Less conveniently, the total eclipse could only be seen from the tropics. So the English astrophysicist Arthur Eddington mounted an expedition to the island of Principe, off the west coast of Africa, while another group set sail for Brazil. The idea was to compare photographs of the Hyades stars during the totality with those taken a few months earlier at night and measure any shifts in the stars’ positions relative to each other. Most of Eddington’s photographs during the eclipse turned out to be useless, because wispy clouds obscured the stars. But one good photograph allowed him to discern a tiny defection: his measurement and Einstein’s prediction were in good agreement. “Through clouds, hopeful,” he telegraphed home. The team in Brazil had better luck with weather, but its photographs could not be examined until the members returned to Europe. On November 6, the official results of the expeditions were presented at a joint meeting of the Royal Society and the Royal Astronomical Society in London. The next day, the Times broke the story with the headline “Revolution in Science—Einstein versus Newton.” Two days later, the New York Times declared, “Lights All Askew in the Heavens—Einstein Theory Triumphs.” The legend of the superstar scientist was born.

Einstein went on to show theoretically that if two stars were to line up exactly, as seen by an observer on Earth, the nearer one would act as a lens magnifying the light from the more distant star, making the latter appear much brighter than usual temporarily. It was an unlikely occurrence, he noted. Later scientists realized that bigger cosmic structures, like galaxies, might indeed line up more often. So since the 1970s, astronomers have identified many examples of gravitational lensing—typically distant quasars whose images are distorted by the gravity of intervening galaxies. But bending of light by individual stars—called microlensing to distinguish it from lensing by entire galaxies containing hundreds of billions of stars—is a lot harder to detect. Since the chance of catching two stars in our own Galaxy in perfect alignment is minuscule, astronomers would have to monitor millions of stars each night to catch a handful of events in progress. But that’s precisely what the Princeton University astronomer Bohdan Paczynski urged his colleagues to do, since the payback could mean valuable insights into what makes up the elusive “dark matter” that holds the Galaxy together. What’s more, in 1991, with then-student Shude Mao, he proposed gravitational microlensing as a means to search for extrasolar planets. His idea was that a planet around the foreground star would alter the magnifying properties of the lens dramatically and thereby betray its presence.

Spurred on by Paczynski’s passionate advocacy, in the early 1990s several research teams commenced surveys for microlensing events. To maximize the odds, they targeted areas of the sky with the highest concentrations of stars, such as the bulge of the Milky Way. Others coordinated networks of small telescopes around the globe to obtain follow-up observations of particularly interesting events revealed by these surveys. The trick was to fag the unusual events early enough and alert the observers in time. While a microlensing event typically unfolds over several weeks, the rapid rise to a peak and the subsequent drop in brightness only lasts a few days.

The e-mail that appeared in Jennie McCormick’s inbox came from Andrew Gould, a professor at Ohio State and leader of the MicroFUN (for Microlensing Follow-Up Network) collaboration. The event he wanted observed, dubbed OGLE-2005-BLG-071, had been detected a month earlier by the Optical Gravitational Lensing Experiment survey team with a 1.3-meter telescope in Chile. Monitoring over the next few weeks had suggested it was likely to be a rare high-magnification event. The amount of magnification depends on how well the two stars are aligned; in the best cases, the result could be a thousandfold increase in brightness. Such spectacular events are ideal for detecting a planet in the lensing star’s midst. Intriguingly, as the time of maximum magnification approached, researchers had noticed that the light curve of OGLE-2005-BLG-071 began to depart from the smooth rise expected from a single lens. That’s when Gould decided to alert as many observatories—both professional and amateur—as he could, to ensure around-the-clock coverage. McCormick, of course, had no idea about all this. The next day, she e-mailed Gould back: “I’ve got data on your target. What should I do with it?” For Gould, her e-mail came out of the blue. He had never heard of her and was skeptical that her observations would be of much use. Not wanting to sound impolite or discouraging, he asked her to send the data over anyway. To his surprise, the data were of superb quality. He asked for more coverage the next night. With clear skies over Auckland, McCormick and Grant Christie, another local amateur using a 14-inch telescope, were able to record the brightness of the target every few minutes for two critical nights near the peak of the event.

The excellent coverage traced strong departures from a simple lens model. That could mean only one thing, according to Gould: “There’s no doubt the lensing star has a planet, which caused the deviation we saw.” Best estimates put the planet at a few times the mass of Jupiter, orbiting a star some 15,000 light-years from Earth, roughly halfway between us and the center of the Galaxy. “It just shows that you can be a mother, you can work full-time, and you can still go out there and find planets,” gushed a proud McCormick, who along with Christie, shared authorship of the scientific paper reporting the discovery in the Astrophysical Journal. For her pivotal contribution, McCormick, who had never before left New Zealand, was rewarded with a trip to Columbus, Ohio, where she spoke to an audience of professional astronomers and graduate students. “Amateurs like her are pretty crucial to what we do. They are the heart and soul of our collaboration,” said Gould’s former student and current colleague Scott Gaudi.

Figure 5.1. How gravitational microlensing reveals an extrasolar planet.

Planet Census

This 2005 discovery was only the second reliable detection of a planet by using microlensing. The first had been announced a year earlier by a large team led by Ian Bond at Edinburgh University. That planet is estimated to weigh about 1.5 Jupiter masses and orbits a red dwarf star 17,000 light-years away, also toward the Galactic bulge. Since 2005, the technique has achieved a degree of maturity, with about a dozen more planets to its credit. The tally includes OGLE-2006-BLG-109, first identified on March 28, 2006, and later reaching a magnification of 500. Once again, the MicroFUN group, including McCormick and Christie, carefully monitored its rise and fall. This time, the complex shape of the light curve implicated not one but two giant planets circling a red dwarf 5,000 light-years away. In fact, this planetary system resembles a scaled-down version of our own, with the inner planet weighing two-thirds as much as Jupiter and the outer one having nearly the same mass as Saturn. While their orbits are smaller than those of Jupiter and Saturn, the temperatures are likely to be similar because their parent star is dimmer and cooler than our Sun. “Theorists have wondered whether gas giants in other solar systems would form the same way as ours did. This system seems to answer in the affirmative,” said Gaudi.

Microlensing has two big advantages over other techniques of searching for planets. One is that it is sensitive to planets as small as the Earth, even if they are a few times farther from their stars than the Earth from its Sun. For instance, in the case of the 2005 event, “if an Earth-mass planet were in the same position, we would have been able to detect it,” explained Gould. In fact, one of the lowest-mass extrasolar planets known to date was found through microlensing. The 5.5-Earth-mass planet, roughly 2.5 AU from a dwarf star some 20,000 light-years away, was announced in 2006 by an international team led by Jean-Philippe Beaulieu at the Institute of Astrophysics in Paris. This “super-Earth”—dubbed OGLE 2005-BLG-390Lb—is almost certainly in a deep freeze, with temperatures approaching that of Pluto, given its distance from the faint parent star.

The other big plus is that by monitoring millions of stars, microlensing can provide an estimate of the frequency of planets in the Galaxy. “With microlensing, we are mapping the demographics of planets,” is how Ohio State’s Gaudi described the technique’s niche. In his PhD thesis, completed in 2000, Gaudi determined that less than one-third of all stars in the Galaxy harbor Jupiter-mass planets at a few times the Earth-Sun separation. With more extensive surveys, it is now possible to estimate the fraction of stars with smaller Neptune-mass planets: that number comes in at about 40 percent, albeit with fairly large uncertainties. “That means Neptunes are the most common type of planets we know of so far,” Gaudi pointed out. The next step in microlensing research is to fold the search phase and follow-up phase into one, through continuous monitoring of a moderate-size patch of the sky toward the Galactic bulge with three 1-to 2-meter-class telescopes spanning the globe. These telescopes will take a picture every twenty minutes. In effect, every event will be followed up. But high-magnification events require more frequent observations. That’s where the amateurs will continue to play an important role. Eventually, Gaudi and others dream of doing supersensitive microlensing surveys from Earth orbit that could detect analogs to all of our solar system’s planets, except the tiniest and innermost Mercury. Their best hope may be to join forces with cosmologists who want to pin down the nature of the mysterious “dark energy” that appears to dominate the universe. Both types of science require a space telescope capable of wide-field imaging.

Microlensing suffers from a huge drawback, however. For all practical purposes, a particular planet host’s perfect alignment with a background star occurs only once, and is never repeated. If astronomers fail to gather enough data during the event, there is then ambiguity about the presence or absence of a planet. In fact, claims of planet detection through microlensing before 2004 suffered from just that problem. What’s more, even if a planet’s signal is definitively identified by microlensing, there is no opportunity for follow-up studies to characterize it—little chance of determining whether it is likely to harbor life, for example.

For detailed studies of individual extrasolar planets, a very different technique has proven best. Here, instead of a temporary magnification due to the chance alignment of two stars, astronomers look for mini-eclipses of a star due to a planet around it passing in front. Unlike microlensing events, these transits occur again and again, each time the planet is in that part of its orbit that intersects our line of sight.

Chasing Silhouettes

In our own solar system, we can see Mercury and Venus, the two planets inside the Earth’s orbit, passing in front of the Sun—but not as often as you might think. That’s because their orbits are tilted just slightly with respect to the Earth’s orbit. In early June and early December every year, Venus intersects the plane of the Earth’s orbit, or the ecliptic, at two nodes that cross the Sun. But a transit is seen only when it happens to pass a node at inferior conjunction—that is, when Venus happens to be exactly between the Earth and the Sun at the same time it crosses the ecliptic. These celestial alignments follow a precise cycle, with time intervals of 8.0, 121.5, 8.0, and 105.5 years in the case of Venus; Mercury transits are more frequent, with about a dozen or so per century. As Venus passes in front of the Sun, taking several hours to do so, it appears as a black dot about one-thirtieth the solar diameter. It’s big enough to be seen with the (properly protected) naked eye, but there are no records of a transit being observed before the invention of the telescope early in the seventeenth century. That’s not too surprising given the rarity of the event.

In 1629 Johannes Kepler, as he investigated the laws of planetary motion, realized that transits of both Venus and Mercury would occur two years later. Unfortunately, he didn’t live to see either, and the Venus transit of 1631 was not visible from Europe in any case. But European astronomers were able to observe the transit of Mercury in November that year, vindicating Kepler’s prediction. Eight years later, Englishmen Jeremiah Horrocks and William Crabtree, friends living thirty miles apart, made the first recorded observations of a Venus transit by projecting the Sun’s image with small telescopes.

Perhaps inspired by a transit of Mercury he observed from the island of St. Helena in 1677, Edmund Halley, of comet fame, presented a paper to the Royal Society in London in 1691 on measuring the distance between the Earth and the Sun—the astronomical unit—using transit timings. His suggestion, an idea also proposed by a Scottish mathematician almost thirty years earlier, was to time the transit from widely separated locations on Earth and use the difference in the apparent paths taken by Venus across the face of the Sun to calculate the Earth-Venus and thus Earth-Sun distance using trigonometry. An accurate measurement of the AU, then known to not much better than a factor of 2, would not only set the distance scale for the solar system but also hold the more down-to-Earth promise of improving celestial tables used for maritime navigation.

After Halley’s death, others took on his charge of organizing expeditions to observe the 1761 and 1769 Venus transits from various parts of the globe. Despite the challenges of long-distance sea travel by wooden sailing ship, the difficulty of obtaining and using precise clocks and other instruments, and the dangers posed by the ongoing Seven Years’ War between England and France, astronomers from those two countries and Austria mounted expeditions to such far-flung venues as Newfoundland, St. Helena, Norway, Siberia, and the Indian Ocean for the first transit. Even with some 120 observers in total, it turned out that the spread in latitude was far from optimal for a parallax measurement. With the added uncertainties of timing, bad weather in some locations, and no knowledge of the exact longitude of others (such as Rodrigues Island just east of Mauritius), the much-hoped-for improvement in measuring the AU did not materialize.

The apparent failures of 1761 made it all the more important to get things right for the second transit of the pair eight years later. A commission set up by the Royal Society called on King George III to support an expedition to Tahiti to observe the 1769 transit. The proposal highlighted the practical value of the result, stating that a transit measurement would “contribute greatly to the improvement of astronomy on which Navigation so much depends.” It even appealed to the national pride: “The French, Spaniards, Danes and Swedes are making the proper dispositions for the Observation thereof. . . . The Empress of Russia has given directions for having the same observed. . . . It would cast dishonor [on the British nation] should they neglect to have the correct observations made of this important phenomenon.” These arguments, and the cost underestimate of 4,000 pounds “exclusive of the expense of the ship,” sound remarkably familiar to today’s scientists asking for government funds for space missions!

The king found the arguments persuasive and approved the funds. The small Royal Navy ship Endeavor set off in August 1768 with Lieutenant James Cook in command. It also carried the English astronomer Charles Green, the Swedish naturalist Daniel Solander, and the gentleman traveler Joseph Banks, who contributed 10,000 pounds of his own money, as well as scientific instruments, crew, and supplies. They arrived in Tahiti six weeks before the transit and set up an observing post at “Point Venus.” The day of the transit “prov’d as favorable to our purpose as we could wish, not a Clowd was to be seen all day and the Air was perfectly clear,” Cook wrote in his journal. But timing the end of ingress and the start of egress, when Venus is just inside the limb of the Sun, proved difficult. Cook’s party noted that at these contact points, the planet’s dark disk appeared distorted in the shape of a raindrop, as if a thread was connecting it to the Sun’s limb. This so-called black-drop effect limited the precision of the transit timings.

After the transit, Cook explored New Zealand and the east coast of Australia, in an attempt to fulfill the second (and secret) part of his mission—to search for a mysterious southern continent. The Endeavor returned to England after three years at sea, and the transit data it brought back from Tahiti, when combined with those from other sites around the world, allowed astronomers to determine the AU to within a few percentage points of its modern value—a remarkable achievement.

I saw a transit of Mercury as a fifteen-year-old in Sri Lanka. But most professional astronomers pay little attention to transits of Mercury and Venus these days. The distance to these planets can now be measured extremely well using radar, and the AU is known to within 30 meters, or about the size of a football field. But transits of another kind—those of planets around other stars—are at the forefront of astrophysics.

Planets from Dips

Stars are so distant that we see them only as points of light. So it’s not possible to track a transiting exoplanet as a dot on a star’s visage. Instead, each time the planet passes in front of it, we measure a tiny dip in the star’s brightness. As seen from a great distance, a planet similar in size to Jupiter would block about 1 percent of the star, causing a periodic dimming of 1 percent, whereas a smaller planet would cover less of the starlight and result in a smaller dip. Thus, the depth of the dip reveals the diameter of the planet, if we know the size of its star. Transit observations are most useful when combined with radial velocity measurements of the star as it wobbles due to the planet’s gravitational tug. Together, they allow astronomers to derive the planet’s average density (mass divided by volume) and therefore infer something about its bulk composition.

Figure 5.2. When a planet transits in front of its star, it covers a small fraction of the star’s visage, resulting in a temporary dip in its brightness. The bigger the planet, the larger the dip.

In the summer of 1999, Tim Brown, then a staff scientist at the High Altitude Observatory, and David Charbonneau, a visiting graduate student from Harvard, set up a small telescope in a parking lot in Boulder, Colorado, to begin a search for transiting exoplanets. With an aperture of just 10 centimeters, their telescope, called STARE (so named for their project, STellar Astrophysics and Research on Exoplanets), was not much bigger than the one Cook’s expedition took to Tahiti. But, with a much bigger field of view and a modern CCD camera, it could monitor thousands of stars at a time for small changes in brightness. Before they were ready for a full-scale transit search, Brown and Charbonneau got an intriguing tip at the end of August from David Latham at the Harvard-Smithsonian Center for Astrophysics. One of the stars whose radial velocities Latham and his collaborators had been measuring appeared to harbor a close-in giant planet. “We knew that our chances [of catching it in transit] were 1/10, assuming the orbit was randomly inclined to our line of sight,” Charbonneau told me. The STARE team sprang into action by early September, taking data on Latham’s target, a Sun-like star 150 light-years away with the catalog number HD 209458. When Charbonneau analyzed the data in October, he found brightness dips characteristic of transits in the data from the nights of September 9 and 16, at exactly the times predicted by Latham’s measurements of the star’s wobble.

Meanwhile, Greg Henry, an astronomer at Tennessee State University with access to an automated telescope in southern Arizona, was also chasing the shadow of HD 209458’s planet. Berkeley professor Geoff Marcy, whose team had detected the star’s wobble from observations at one of the two Keck telescopes in Hawaii, independently of Latham, had alerted him to it. Henry started observing the star on the night of November 7 and caught the first part of the dip, called the ingress, as the planet started to cross the star. Unfortunately he couldn’t follow the star long enough to see the end of the transit, because the star was too low in the sky by then. Nevertheless, given the excellent match between the transit time predicted from Marcy’s velocity measurements and Henry’s detection of a partial transit, the team announced its discovery in an International Astronomical Union Circular on November 12 and issued a press release a couple of days later.

Understandably, the STARE team was not pleased. There had been a few delays in submitting its results for publication. Meanwhile, the other team had rushed to make the announcement of a partial transit. In the end, however, both teams’ papers, submitted on the same day to Astrophysical Journal Letters, were published in the same January 20, 2000 issue. It often happens in science that two or more independent groups of researchers hit upon the same quarry at almost the same time. In this instance, unlike in some others, both teams received due recognition from their peers.

By 1999, only a few hardliners questioned the reality of extrasolar planets revealed by the Doppler method. The discovery of HD 20945 8b’s transits put any lingering doubts to rest. What’s more, from the depth of the transit, astronomers could infer the planet’s radius and confirm that it is indeed a gas giant.

Yet, this first-discovered transiting planet remains one of the oddest. It weighs barely two-thirds as much as Jupiter, but is 35 percent bigger. That means its average density is only about one-third that of water and about one-half that of Saturn, the least dense world in our solar system. In other words, HD 209458b is extremely bloated! Gas giant planets are born big and hot, but they contract and cool down over time. Given that its parent star is several billion years old, HD 209458b should have shrunk to roughly the same size as Jupiter. Why hasn’t it? One obvious difference is that it is much closer to its star than Jupiter is to the Sun, thus it receives a lot more stellar heat. But theorists’ calculations suggest that the star’s radiation is far from sufficient to keep the planet inflated. There must be an extra source of heat. One possibility is that tides are at work. Just as the Moon raises tides in the Earth’s oceans, stars will raise tides in the atmospheres of close-in giant planets. In most cases, the star tugging on the tidal bulge over millions of years would cause the planet to end up in a circular orbit with its orbital plane aligned with the star’s equator. The orbit of HD 209458b is indeed circular, as expected. But its orbit could be tipped by as much as 4 degrees. In that case, the tidal bulge will slosh north and south as the planet orbits the star, generating internal heat to keep the planet bloated. Still, there are theorists who wonder whether tidal heating is enough to account for the unusually large size of HD 209458b.

Figure 5.3. The transit of HD 209458b observed with the Hubble Space Telescope. Credit: T. Brown (Las Cumbres Observatory Global Telescope) et al./NASA

When I attended a conference on extrasolar planets in Washington, DC, in the summer of 2002, transits were all the rage. Keith Horne from the University of St. Andrews counted two-dozen transit searches in the works, employing a variety of instruments ranging from wide-field cameras that used commercially available 200-millimeter Canon lenses to the majestic 4-meter telescope on Cerro Tololo, Chile. But, by summer 2002, almost three years after HD 209458b was confirmed, no other discoveries had been reported.

The second transiting exoplanet, and the first one to be found rather than confirmed with this method, came six months later from an unexpected corner: a survey for gravitational microlensing events. A team led by Harvard professor Dimitar Sasselov examined fifty-nine stars toward the Galactic bulge identified by the OGLE project as having brightness dips resembling planetary transits. Spectroscopic follow-up with telescopes in Arizona and Chile revealed most of them to be binary stars that undergo grazing eclipses or contain a faint stellar companion. But five promising candidates remained. With more intensive spectroscopic observations at Keck in Hawaii, Sasselov’s team confirmed that one star—dubbed OGLE TR-56 and located 5,000 light-years away—indeed harbors a hot Jupiter. The planet is so close to the star, barely four times the star’s radius away, that it completes a “year” every twenty-nine Earth-hours—setting a new record and posing a new puzzle as to what stopped the planet from falling all the way into the star.

By 2010, nearly 100 transiting exoplanets had been identified. Most are gas giants akin to Jupiter and inhabit tight orbits, completing a “year” every few Earth-days. That is because bigger planets cause bigger dips, making them easier to detect, and because planets closer to their parent stars are more likely to line up against the stellar backdrop. The SuperWASP (Wide Angle Search for Planets) project, headed by astronomers in the United Kingdom, is one of the most prolific, with over forty discoveries to its credit. The project uses eight automated wide-field cameras at each of two sites—one in the Canary Islands and the other in South Africa—for the search and a variety of bigger telescopes around the world for confirmation with radial velocity measurements. Another team, led by Kailash Sahu of the Space Telescope Science Institute in Baltimore, found sixteen transiting planet candidates using the Hubble Space Telescope. The strategy was to take 519 pictures of nearly a quarter-million stars toward the Milky Way’s bulge over seven straight days in 2004. After processing the images, the team looked for tiny brightness dips characteristic of planet transits, taking care to eliminate others that were due to grazing eclipses among binary and triple star systems. Unfortunately, of the sixteen good candidates they announced in 2006, only two are around stars bright enough for measuring precise velocities with spectra for confirmation. Five of Sahu’s planet candidates are truly extreme worlds: with orbital periods shorter than 1.2 days, they must be seething under the parent stars’ heat and are probably stretched into egg shapes by strong tidal forces.

Another strange beast in the zoo of transiting planets is a “hot Saturn” that circles the yellow subgiant star HD 149026 every three days. This planet, found in 2005, is so extraordinarily dense that one-half to two-thirds of its mass must consist of heavy elements—comparable to the total in our solar system’s planets combined. One model suggests that it has a solid core, nearly seventy times the mass of the Earth, surrounded by a layer of super-dense water under a mantle of liquid metallic hydrogen. Its origin remains a mystery. One idea is that its core grew over time after it had migrated close to the star, through the accumulation of solid particles from the protoplanetary disk while the star stripped it of its gaseous outer layers. In another, more dramatic scenario, HD 149026b was formed through the collision and merger of two conventional protoplanets, each about thirty-five times the Earth’s mass.

Figure 5.4. Jupiter is thought to have a solid core 5–15 times the mass of the Earth, whereas the core of HD 149026b is probably much bigger, perhaps as hefty as 70 Earth masses. Credit: G. Laughlin (UCSC)/oklo.org

Two of the smallest exoplanets seen in transit are just slightly bigger than Neptune. The first, around the red dwarf star Gliese 436, was identified in a radial velocity survey by the California-Carnegie team and later seen in transit. The other, HAT-P-11b, was found in a transit search carried out by Harvard-Smithsonian astronomers using small, automated telescopes in Arizona and Hawaii. These planets cover less than 0.5 percent of their star’s light, making the detection of their transits challenging for ground-based telescopes. But astronomers can find even smaller worlds through transits by using satellite observatories or targeting dwarf stars, as we will see in chapter 9.

Weather Reports

Transits not only betray a planet’s presence and reveal its orbital period, size, and bulk composition (the latter when combined with velocity measurements) but also offer other interesting prospects. For example, if a planet has a ring system, like Saturn’s, it could result in tiny dips just before and right after the main brightness drop due to the planet itself. In fact, the rings of Uranus were discovered serendipitously in a manner similar to this. Just by chance, Uranus was to line up exactly with a fairly bright star on March 10, 1977. Three astronomers took observations during this “occultation” with a telescope on a plane called the Kuiper Airborne Observatory, hoping to study the planet’s atmosphere as some of the distant star’s light passed through it on the way to us. When they analyzed the data later, the researchers noticed that the star had disappeared briefly from view five times both before and after the planet eclipsed it. The reason, they deduced, was the presence of narrow rings around Uranus. If the star had dimmed only on one side of the main eclipse, a moon rather than a ring could have been responsible. So far, astronomers observing transiting exoplanets have not seen evidence of rings or big moons, but they have not given up looking.

Scientists have successfully used transits to probe the atmospheres of alien worlds. That’s because when a planet passes in front of its star, a bit of the starlight skims the planet’s upper atmosphere before reaching us. Imprinted on that light, as weak absorption lines, are the telltale signatures of various gases. By comparing spectra of the star taken while the planet is in and out of transit, astronomers can identify chemicals in the planet’s atmosphere. Soon after the detection of HD 209458b in transit in 1999, Sara Seager, then at the Institute for Advanced Study in Princeton, and Dimitar Sasselov of Harvard wrote a theoretical paper predicting which chemical species would be easiest to spot. Given the extreme temperatures of hot Jupiter planets, they suggested that water vapor, carbon monoxide, alkali metals, and possibly methane should leave detectable imprints. Of these, the alkali metals sodium and potassium absorb light in the visible part of the spectrum.

Just two years later, a team led by Charbonneau and Brown detected sodium in the atmosphere of HD 209458b using the Hubble space telescope, a breakthrough that could someday lead us toward unveiling biosignature molecules in the atmospheres of terrestrial worlds. For now, their discovery confirmed our basic picture of hot Jupiters. But the sodium absorption was a lot weaker than theorists had expected. The most likely explanation is that high clouds are blocking our view of sodium in the atmosphere below. Using the same subtraction method at ultraviolet wavelengths, a French team has reported a large comet-like halo of gas, mostly hydrogen, around HD 209458b. The UV radiation from the star must be heating these molecules and enabling them to escape from the atmosphere. Astronomers continue to debate just how much of the planet’s atmosphere has been lost over its lifetime. In 2010, researchers using the Hubble found evidence of gas escaping from another hot Jupiter, WASP-12b. This planet is so close to its parent star that strong tidal forces may have stretched it into the shape of an American football, according to theoretical calculations by Douglas Lin and his collaborators. Its dayside temperature may be as high as 2600 degrees Celsius.

The going is much tougher using ground-based telescopes because of contamination by the Earth’s own atmosphere, and several groups failed in their attempts to detect species like methane and sodium in hot Jupiter atmospheres. Seth Redfield, now at Wesleyan University, and colleagues sighted sodium at last in 2007 in HD 189733b, with the 9.2-meter Hobby-Eberly Telescope in west Texas. The sodium imprint is three times stronger in this planet than in HD 209458b, indicating differences between their atmospheres: the latter appears to have a high cloud deck while the former doesn’t. In 2010, a team led by Ignas Snellen of Leiden University used the Very Large Telescope in Chile to measure carbon monoxide in the atmosphere of HD 209458b; they found evidence for a “super wind,” blowing at thousands of kilometers per hour. Being able to unravel that kind of detail about planets we can’t take pictures of from where we are, tens of light-years away, is pretty remarkable.

Because of their close proximity to the stars, hot Jupiters are expected to be tidally locked, with one side of the planet always facing its sun, just as one hemisphere of the Moon always points toward the Earth. The big question is whether strong winds on these planets spread the heat from the permanent dayside to the eternal nightside. If they do, both hemispheres will have similar temperatures. Otherwise, one side will be scorching hot while the other side endures everlasting chills.

A planet that transits in front of its star also slips behind the star for part of its orbit (except in very rare cases). Just before a planet goes into such a “secondary eclipse,” its dayside, fully illuminated by starlight like the full Moon is by the Sun, is facing us. Astronomers have exploited this fact to detect the infrared emission—or heat—from several exoplanets directly. Like many other measurements in astronomy, this one is done in a relative sense: astronomers measure the light from the combined star and planet just before the secondary eclipse and then subtract from it the light from the star alone when the planet is hidden behind. What’s left is the planet’s feeble emission, from which we can deduce its dayside temperature. In 2005, two teams of astronomers —one led by Drake Demming of NASA’s Goddard Space Flight Center and the other by Harvard’s David Charbonneau—did just that for two exoplanets, HD 209458b and TrES-1b, using the Spitzer space telescope. Both planets have temperatures of about 830 degrees Celsius, much as we might expect for these hot Jupiters broiling in their star’s glare.

Heather Knutson, who did her PhD with Charbonneau at Harvard, has gone further, making a crude map of an exoplanet’s heat distribution. Knutson grew up in the Marshall Islands, a small coral atoll in the Pacific about halfway between Hawaii and Australia. Her parents worked on the U.S. Army missile range on the island of Kwajalein, whose isolated location near the equator made for excellent stargazing. As a child, Knut-son often ventured out to the edge of the island armed with a red flashlight and a book of constellations.

She and her colleagues stared at the hot Jupiter HD 189733b for half its orbit—thirty-three hours—with the Spitzer space telescope. During the primary transit, Spitzer would be looking at the planet’s dark side, but as it continued in its orbit, more of its dayside rotated into view, with the entire bright half visible just before the planet went into the secondary eclipse behind the star. As a result, the scientists were able to make a simple map of how temperature varies with longitude on an exoplanet for the first time. The map revealed a single hot spot that is about twice as big as the Great Red Spot on Jupiter and much hotter, at 940 degrees Celsius. Interestingly, the hottest point on the planet is not at “high noon”—that is, directly under its sun—but is offset in longitude by about 30 degrees. That is probably the result of strong winds redistributing heat within the planet’s atmosphere. The same winds appear to carry heat over to the nightside, making even the coldest regions a balmy 700 degrees Celsius. “Even the nights are steaming hot on this world,” said Knutson. “This planet has powerful jet streams,” added Charbonneau. “While the Earth’s jet stream blows at about 300 kilometers per hour, the jet stream on HD 189733b may blow as fast as 9,000 kilometers per hour, according to computer models.”

Since then, astronomers have managed to take the temperature of several other exoplanets, using Spitzer and Hubble at first, and more recently with ground-based telescopes. Bryce Croll, a graduate student working with me at the University of Toronto, is among those chasing secondary eclipses. So far, we have detected the eclipses of four of the hottest exoplanets, using the 3.6-meter Canada-France-Hawaii telescope on Mauna Kea. Ours are the most precise measurements yet of planetary eclipses from the ground. For two of the planets we observed, the temperatures appear to be similar on the permanent dayside and the nightside. “Since the night sides of these planets never see the star, this is as much of a surprise as finding that the temperature at the Earth’s north pole in the middle of winter is the same as at the equator,” Croll pointed out. For the other two, with dayside temperatures hot enough to melt iron and even platinum, there is a likely difference of several hundred degrees, suggesting that they do not harbor winds strong enough to spread the heat around. That’s also the case with upsilon Andromedae b,¹ a planet observed with Spitzer. It is fiery hot on one side and icy cold on the other—a difference of 1400 degrees Celsius. “If you were moving across this planet from the nightside to dayside, the temperature jump would be equivalent to leaping into a volcano,” said Brad Hansen of the University of California at Los Angeles, one of the scientists who measured its temperature variations.

If you think that’s extreme, consider the wild temperature swings on HD 80606b. A gas giant a few times more massive than Jupiter orbiting a Sun-like star, it was discovered by the Swiss planet hunters back in 2001. Its 111-day orbit, they found, is highly elongated, more like a comet’s path than a planet’s. At one end of its orbit, the planet is almost as far from the star as the Earth-Sun distance, while at the other it goes in closer than our Mercury. “If you could float above the clouds of this planet as it approaches the innermost point of its orbit, you’d see its sun growing larger and larger at faster and faster rates, increasing in brightness by almost a thousandfold,” said Greg Laughlin of the University of California at Santa Cruz. In late 2007, he and his collaborators used Spitzer to observe HD 80606b before, during, and just after its closest encounter with the star. What they saw was nothing short of dramatic: the planet heated up—by 700 degrees Celsius—and then cooled back down in a matter of hours. Such extreme temperature swings no doubt give rise to fierce storms on this alien world. “This is the first time that we’ve detected weather changes in real time on a planet outside our solar system,” Laughlin added.

Weather reports from other extrasolar worlds are arriving now. And, as the Spitzer mission winds down, astronomers look forward to using the Stratospheric Observatory for Infrared Astronomy (SOFIA), a 2.5-meter telescope mounted on a Boeing 747. At an altitude of 12 kilometers, it rises above most of the water vapor in the Earth’s atmosphere, which otherwise hampers long-wavelength observations. If funding permits, SOFIA may fly three or four nights a week starting in 2010 (it had the first fight in May 2010, but regular operations start in 2011) for ten years or more, out of Edwards Air Force Base in California. It will build upon Spitzer’s legacy of characterizing exoplanets, while also studying a variety of other astronomical objects—from star-forming clouds in our cosmic backyard to newborn galaxies in the distant universe.

The James Webb Space Telescope, a 6.5-meter successor to Hubble operating at near-and mid-infrared portions of the spectrum, is the next big thing on the horizon. Scheduled for launch in 2014, it should be able to sense the heat from hot Neptunes and possibly even big terrestrial planets. The next time you hear about “the storm of the century,” it may well be from astronomers rather than meteorologists, talking about violent weather raging on a steaming world circling a distant solar twin.

¹ This hot Jupiter doesn’t transit its star, but astronomers using the Spitzer space telescope were still able to measure the tiny changes in its brightness as it showed different phases at different points in the orbit.